Protodecarboxylation of carboxylic acids over heterogeneous silver catalysts

Article abstract of DOI:10.1039/c3cy00580a

A heterogeneous supported Ag catalyst for the protodecarboxylation of aromatic carboxylic acids has been developed. Control of the metal particle size proved extremely important. The highest activity was achieved with a silver loading of 10 wt%, which had relatively big metal crystallites of 40 nm. It is inferred that the adsorption of the aromatic moiety requires terrace sites rather than edges or corners at the metal nanoparticle. The amphoteric support, γ-Al₂O₃, gave the most active catalysts. Oxygen coverage of the surface is essential for catalytic activity. A mechanism has been proposed with the critical steps (1) formation of a benzoyl anion by reaction with a base in the reaction medium, (2) binding of the anionic species at the Ag⁺ surface sites with (3) extrusion of CO₂ and (4) proton transfer from another molecule of carboxylic acid, followed by desorption of the decarboxylated species and binding of the benzoate to the active site to complete the catalytic cycle. With 2-nitrobenzoic acid as substrate, the catalyst had a turnover frequency (TOF) of 216 h^-1. The catalyst showed good activity for benzoic acid with nitro, methoxy and halogen substituents at the ortho-position as well as for heteroaromatic carboxylic acids. The Royal Society of Chemistry.

Full text of DOI:10.1039/c3cy00580a

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Protodecarboxylation of carboxylic acids
over heterogeneous silver catalysts†
Cite this: Catal. Sci. Technol., 2014,
4, 516
Xiu Yi Toy, Irwan Iskandar Bin Roslan, Gaik Khuan Chuah and Stephan Jaenicke*
A heterogeneous supported Ag catalyst for the protodecarboxylation of aromatic carboxylic acids has
been developed. Control of the metal particle size proved extremely important. The highest activity was
achieved with a silver loading of 10 wt%, which had relatively big metal crystallites of 40 nm. It is inferred
that the adsorption of the aromatic moiety requires terrace sites rather than edges or corners at the
2 3
metal nanoparticle. The amphoteric support, γ-Al O , gave the most active catalysts. Oxygen coverage of
the surface is essential for catalytic activity. A mechanism has been proposed with the critical steps (1)
formation of a benzoyl anion by reaction with a base in the reaction medium, (2) binding of the anionic
+
Received 6th August 2013,
Accepted 30th October 2013
species at the Ag surface sites with (3) extrusion of CO2 and (4) proton transfer from another molecule
of carboxylic acid, followed by desorption of the decarboxylated species and binding of the benzoate to
the active site to complete the catalytic cycle. With 2-nitrobenzoic acid as substrate, the catalyst had a
DOI: 10.1039/c3cy00580a
−1
turnover frequency (TOF) of 216 h . The catalyst showed good activity for benzoic acid with nitro,
www.rsc.org/catalysis
methoxy and halogen substituents at the ortho-position as well as for heteroaromatic carboxylic acids.
as only 10 mol% of the catalyst was required and decar-
boxylation of para-substituted aromatic carboxylic acids could
1
. Introduction
Carboxylic acids are very attractive intermediates as they are
readily available in a wide range of structures and at low cost.
be carried out at 170 °C instead of temperatures above 200 °C.
1
With microwave heating, the reaction time could be shortened
9
Carboxylate groups can direct the regioselectivity in aromatic
substitution reactions. Over the years, several catalytic meth-
to 5–15 min. Decarboxylative methodologies have also been
2
used in the presence of D O to incorporate deuterium
2
,3
10
odologies involving carboxylic acids have been developed.
selectively into aromatic rings. This gives access to valuable
deuterium labelled compounds.
Protodecarboxylation, where surplus carboxylate groups are
removed and replaced with hydrogen atoms, is an important
reaction within this strategy. One of the earliest reports of a
Besides the copper-based catalysts, precious metals were
also reported to be active for protodecarboxylation of aromatic
1
1
catalytic decarboxylation reaction was by Shepard et al. in
carboxylic acids. Dickstein et al. used Pd(CF COO) in the
3 2
4
1
930. These authors obtained halogen-substituted furans by
decarboxylation of 2,6-disubstituted benzoic acids. Although
the reaction temperature was much lower, 70 °C, the substrates
were limited to only benzoic acids with two ortho substituents,
decarboxylating halogenated furancarboxylic acids with cop-
per bronze as catalyst and a high boiling point base, quino-
line. Nilsson's group decarboxylated 2-nitrobenzoic acid using
and 10 equivalents of trifluoroacetic acid were needed as addi-
5
12
a stoichiometric amount of copper(I) oxide. The use of chelat-
tive. Sun and Zhao found that [(cod)Rh(μ-OH)] gave yields of
2
ing agents such as 1,10-phenanthroline and 2,2′-bipyridine
extended the scope of the reaction to a wider range of acti-
vated carboxylic acids such as benzoic acids with electron-
withdrawing ortho substituents and heteroaromatic carboxylic
71–99% for the decarboxylation of a small range of fluorinated
benzoic acids. However, the use of an expensive Rh-based cata-
lyst and the need for phosphine ligands to fine-tune its activity
are drawbacks of this methodology. Subsequently, Nolan's
6
,7
acids. In 2007, Gooßen and coworkers reported an excellent
group described a strongly basic N-heterocyclic carbene
yield of 82% for the decarboxylation of 4-methoxybenzoic acid
(NHC)–gold complex for decarboxylation reactions in the pres-
ence of toluene at 110 °C. Larrosa and coworkers reported
8
13
with the Cu O–phenanthroline–quinoline system. This was a
2
significant improvement over previously reported methodologies
6 4 2 2
the use of o-Ph-C H (t-Bu) PAuCl, together with Ag O as addi-
tive and DMF as solvent, to mediate the decarboxylation of aro-
matic and heteroaromatic carboxylic acids. By modulating
the reactivity of the gold(I) species via ligands, the reaction
could be carried out at temperatures as low as 60 °C. However,
the gold(I)–aryl compounds obtained upon decarboxylation did
Department of Chemistry, National University of Singapore, Kent Ridge,
Singapore 117543, Singapore. E-mail: chmsj@nus.edu.sg;
Fax: +65 6779 1691; Tel: +65 6516 2918
14
†
Electronic supplementary information (ESI) available: Fig. S1–S6, Table S1,
estimation of surface Ag atoms. See DOI: 10.1039/c3cy00580a
516 | Catal. Sci. Technol., 2014, 4, 516–523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
not undergo protodemetallation. Due to the formation of these
gold(I)–aryl species, a stoichiometric amount of the expensive
Au(I)-based ‘catalysts’ was required.
addition, commercial Ag powder (Alfa Aesar, spherical silver
particles, 365 mesh) and Ag O (Alfa Aesar, 99+%) were also
2
used. These materials are referred to as Ag (commercial) and
Silver acetate together with a basic additive, K CO , was
Ag O (commercial).
2
2
3
found to effectively decarboxylate a wide range of ortho-
2
.3 Characterization of alumina supported silver catalyst
substituted benzoic acids at a relatively low reaction tempera-
1
5
ture of 120 °C. The protodecarboxylation of ortho-substituted
aromatic carboxylic acids by Ag CO could be extended to
Powder X-ray diffraction patterns were measured using a
Siemens D5005 diffractometer equipped with a Cu anode and
variable primary and secondary beam slits (scanning rate
2
3
heteroaromatic carboxylic acids by the addition of acetic
16,17
−
1
acid.
These systems are preferred as they are simple silver
0.02° s , dwell time 1 s). The crystallite size of the silver parti-
salts without the need for ligands. However, their separation
and recovery present problems as they are soluble in the
heated reaction mixture. This can be overcome with heteroge-
neous catalysts which can be easily recovered after reaction,
cles was estimated from the full width at half maximum
2
5
(
FWHM) of the Ag (111) peak using the Scherrer equation.
Nitrogen adsorption and desorption isotherms were obtained
at −196 °C with a TriStar 3000 (Micromeritics) porosimeter.
The samples were degassed at 300 °C for 5 h prior to measure-
ments. The specific surface area was calculated using the BET
18
regenerated and reused. In addition, surface properties such
as acidity can be fine-tuned by an appropriate choice of the
1
9,20
support to control activity and selectivity.
For instance, het-
(
Brunauer–Emmett–Teller) equation. Pore size distributions
erogeneous supported silver catalysts have been studied exten-
were calculated from the desorption branch of the isotherm
by the BJH (Barrett–Joyner–Halenda) method. X-ray photoelec-
tron spectra were measured on a VG Mark II Escalab using
monochromated Mg Kα radiation (1253.6 eV) and a hemi-
spherical electron analyzer at a constant transmission energy
pass (50 eV). The binding energies were calibrated with refer-
ence to the C 1s peak at 284.6 eV. The relative surface concen-
tration of Ag to Al was deduced from the Ag 3d5/2 and Al 3p3/2
peaks after correction using the atomic sensitivity factors
provided by the manufacturer (ASFAg = 5.198, ASFAl = 0.25).
Transmission electron micrographs were obtained using a
JEOL 2010. 5–10 mg of the powder sample were suspended in
2
1
22
sively and used in reactions such as oxidation, reduction
2
3
and N-alkylation. Thus, in the present study, we developed
heterogeneous silver catalysts for the protodecarboxylation
using 2-nitrobenzoic acid as a model compound (Scheme 1).
The protodecarboxylation of other aromatic carboxylic acids
was also investigated.
2
. Experimental
All reagents were purchased from commercial sources and
were used as received. Solvents were dried over 4 Å molecular
sieves before use.
2-propanol and sonicated for 15 min. A drop of the resultant
slurry was deposited on a copper grid and dried overnight
under ambient temperature before measurements.
2
.1 Preparation of supported silver catalysts
Catalysts with 10 wt% silver loading on the supports γ-Al
2 3
O ,
2.4 Catalytic studies
2
4
MgO, ZnO , SiO , TiO were prepared by wet impregnation. In
2
2
2
The catalytic testing was carried out in a round-bottomed flask
equipped with a Liebig condenser. Typically, 2-nitrobenzoic
a typical preparation of 10 wt% Ag/Al O , 0.1570 g (0.9270 mmol)
2
3
3
of AgNO was dissolved in 15 ml of deionised water and 0.9 g
acid (2 mmol, 0.3342 g) and K CO (0.3 mmol, 0.0415 g) were
of γ-Al O was added. The mixture was stirred at 70 °C until
2
3
2
3
dissolved in 4 mL of N-methylpyrrolidone (NMP) as solvent
and heated to 120 °C. After adding 10 mol% of the supported
silver catalyst (0.2157 g catalyst containing 21.6 mg, 0.2 mmol Ag),
the reaction was conducted under vigorous stirring. Aliquots
were removed from the reaction mixture, filtered using a
syringe filter and diluted 50 times with NMP. The diluted
reaction mixture was then analyzed by HPLC (Shimadzu
LC-10AT) with the UV detector set at λ = 254 nm. A reversed
phase C-18 Shim-pack VP-ODS column (Shimadzu) was used
as the stationary phase and the eluent, 50 : 50 : 0.4 (v/v/v) iso-
propanol : water : acetic acid, was delivered isocratically at a
almost dry. The slurry was dried in an oven at 100 °C over-
−
1
night and calcined in air at 500 °C (heating rate 1 °C min )
for 1 h. In addition, catalysts with a silver loading of 5, 10, 15,
and 20 wt% on γ-Al O were prepared using the same method.
2
3
2
.2 Preparation of unsupported silver catalysts
Unsupported silver and silver(I) oxide was prepared by ther-
mal decomposition of commercial Ag CO at a temperature
2
3
of 400 °C and 200 °C for 1 h, respectively. The products are
referred to as Ag (ex-Ag CO ) and Ag O (ex-Ag CO ). In
2
3
2
2
3
−
1
flow rate of 0.5 mL min . Calibration curves were obtained
for 2-nitrobenzoic acid and nitrobenzene.
3
. Results and discussion
3.1 Catalyst characterization
The γ-Al O and SiO -supported silver catalysts had high sur-
face areas of 130 and 369 m g , respectively (Table 1). In
Scheme 1 Protodecarboxylation of 2-nitrobenzoic acid using supported
Ag catalyst.
2
3
2
2
−1
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 516–523 | 517

View Article Online
Paper
Catalysis Science & Technology
Table 1 BET surface area, pore volume and Ag crystallite size of the
supported silver catalysts
the binding energy (BE) and kinetic energy (KE) are of the
same magnitude but in opposite directions. The α′ value
28
lies between 722.68 and 723.99 eV (Table 2), which is in the
Surface area
Pore volume Ag crystallite
2
−1
3
−1
a
29,30
Catalyst
(m g
)
(cm g
)
size (nm)
range of Ag O, while metallic Ag is expected at ~726 eV.
2
The formation of an oxidized surface overlayer is not unex-
pected as the samples have been calcined in air at 500 °C.
However, as observed from the XRD spectra, the bulk of the
silver is in the metallic form. The Ag/Al ratio as deduced by
XPS is higher than that expected from the loading. This is
due to the method of catalyst preparation where silver was
impregnated onto the Al O support, resulting in a high con-
1
1
1
1
5
0 wt% Ag/ZnO
0 wt% Ag/TiO
8.2
36
90
369
145
130
118
114
148
0.2
0.05
0.12
0.14
1.01
0.27
0.25
0.22
0.21
0.27
114
72
63
27
5
40
66
93
n.a.
—
—
—
—
2
0 wt% Ag/MgO
0 wt% Ag/SiO
wt% Ag/γ-Al
0 wt% Ag/γ-Al
5 wt% Ag/γ-Al
2
2 3
O
1
1
2
O
3
2
O
2
O
3
20 wt% Ag/γ-Al
3
2
3
2 3
γ-Al O
Ag (ex-Ag
Ag (commercial)
Ag
Ag
b
centration of the metal at the surface of the catalyst particles.
2 3
CO )
0.003
<0.001
0.003
0.02
0.9
0.3
c
2
O (ex-Ag
2
CO
3
)
2
O (commercial)
<0.001
3.2 Catalytic studies
a
c
b
3
.2.1 Optimization of reaction conditions. Silver supported
From Scherrer equation. Calcining Ag
2
CO
3
at 400 °C for 1 h.
2
Calcining Ag CO
3
at 200 °C for 1 h.
on ZnO or MgO (Table 3, entries 2 & 3) was inactive for the
protodecarboxylation of 2-nitrobenzoic acid to nitrobenzene.
In contrast, when silver was supported on SiO , TiO or Al O ,
2
2
2 3
contrast, the surface areas of the 10 wt% Ag/ZnO and 10 wt%
the conversion was between 65 and 81% after 16 h (Table 3,
entries 4–6). No conversion was observed over the alumina
support alone.
2
−1
Ag/TiO
show the reflexes of Ag as well as those of the support. Due to
the amorphous nature of γ-Al and SiO , broad peaks were
2
were below 40 m g . The powder X-ray diffractograms
O
2 3
2
Using 10 wt% Ag/Al
2 3
O as the standard catalyst, the reac-
present, while sharp narrow lines were observed for crystalline
tion was optimized by varying the additives and solvent used.
MgO, TiO and ZnO (Fig. S1†). Using the Scherrer equation,
2
the silver crystallites were estimated to be ~27 and 40 nm on
2 3
Table 2 XPS results for Ag/Al O
2 2 3
SiO and γ-Al O , respectively. On supports that had a smaller
specific surface area, the silver agglomerated and the particles
grew to a size of 63 to 114 nm.
BE Ag
3d5/2
(eV)
KE Ag
Ag/Al
XPS
Ag
M N
4 45
N
45
α′-Parameter
(eV)
(wt%)
(eV)
Expected
2 3
As the silver loading on the γ-Al O increased, both the
5
1
1
367.05
367.20
367.25
367.40
356.21
356.54
355.43
356.59
723.26
723.74
722.68
723.99
0.04
0.09
0.14
0.18
0.03
0.05
0.08
0.12
surface area and pore volume decreased (Table 1). However,
even at 20 wt% Ag, the sample still had a surface area of
0
5
2
−1
1
14 m g . The pore size distribution of the samples was
20
rather wide, ranging from around 3 to 7 nm, with the average
pore diameter at 4.6 nm. Despite increasing the silver load-
ing, the pore size distribution of the supported samples
remained similar to that of the γ-Al O support (Fig. S2†). Dif-
Table 3 Influence of support, base and solvent on activity
Entry Catalyst Solvent Additive Temp. (°C) Conv. (%)
2
3
fraction peaks of metallic silver were observed in the powder
X-ray diffractograms (XRD) of the Al -supported samples
1
2
3
4
5
6
7
8
9
10
Al
2
O
3
NMP
NMP
NMP
NMP
NMP
NMP
NMP
NMP
DMF
DMSO
DMA
Dioxane
Toluene
DMA
DMA
DMA
DMA
DMA
K
K
K
K
K
K
2
2
2
2
2
2
CO
CO
CO
CO
CO
CO
3
3
3
3
3
3
120
120
120
120
120
120
120
120
120
120
120
101
111
120
120
120
120
120
120
0
0
0
80
65
81
5.9
5.6
38
76
93
5.0
20
86
93
7.5
23
0.9
35
2 3
O
10 wt% Ag/ZnO
10 wt% Ag/MgO
10 wt% Ag/SiO
10 wt% Ag/TiO
10 wt% Ag/Al
10 wt% Ag/Al
10 wt% Ag/Al
with 5 to 20 wt% loading (Fig. S3†). The crystallite size
increased with loading from 5 to 93 nm. From transmission
electron microscopy (TEM), the metal particles showed a
broad size distribution with the average crystallite size being
smaller than that obtained from XRD (Fig. S4†). This discrepancy
could be due to the smaller sampling size in TEM than in XRD.
To examine the oxidation state of surface silver atoms on
the supported catalysts, XPS line positions for the 5–20 wt%
2
2
2
2
2
O
O
O
3
3
3
3
3
3
3
3
3
3
—
AcOH
K2CO3
10 wt% Ag/Al O
2
10 wt% Ag/Al
10 wt% Ag/Al
10 wt% Ag/Al
10 wt% Ag/Al
2
2
2
2
O
O
O
O
K
2
K
2
K
2
K
2
CO
CO
CO
CO
3
3
3
3
1
1
1
1
2
3
2 3
Ag/Al O catalysts were determined (Fig. S5†). The difference
a
b
in the chemical shift for Ag(0) and Ag(I) is small and not
unambiguous. Furthermore, with sample charging, it is diffi-
cult to accurately distinguish between the different oxidation
14
10 wt% Ag/Al O
K2CO3
2
1
1
1
1
5
6
7
8
10 wt% Ag/Al
Ag (ex-Ag CO
Ag O (ex-Ag CO
Ag (commercial)
Ag
2
O
)
K
2
K
2
K
2
K
2
K
2
CO
CO
CO
CO
CO
3
3
3
3
3
2
3
2
2
3
)
2
6,27
states of silver.
which is the sum of the binding energy of the Ag 3d5/2 photo-
electron and the kinetic energy of the Ag M VV Auger
Hence, the modified Auger parameter, α′,
19
2
O (commercial) DMA
4
Reaction conditions: 2 mmol 2-nitrobenzoic acid, catalyst (0.2 mmol Ag),
electron, is used instead. The value of α′ is independent of
any chemical shifts due to surface charging as the shifts in
a
1
5 mol% K
2
2 3
CO , 4 mL solvent, 120 °C, 16 h. Pretreatment in flowing
b
H at 300 °C for 0.5 h before use. Recycled catalyst.
518 | Catal. Sci. Technol., 2014, 4, 516–523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
The presence of a base, e.g., K CO , was necessary to obtain a
high conversion. Without the base, the conversion remained
below 6% (Table 3, entries 7 & 8).
silver loading of 5 and 10 wt%, respectively (Table S1†).
Beyond this loading, the rate fell.
2
3
To elucidate the efficiency of the supported silver catalyst
as a function of metal loading, the turnover frequency (TOF)
was calculated as function of the number of external surface
atoms available for reaction, Ns,rxn (ESI†). This number was
Different solvents were tested for the reaction, and polar
aprotic solvents were found to favour the reaction (Table 3,
entries 9–13). In solvents such as dimethyl sulfoxide (DMSO),
dimethylacetamide (DMA) and NMP, conversions of 76–93%
were obtained. The lone pairs of electrons on the oxygen,
nitrogen or sulfur atoms allow these solvents to form σ-bond
with the silver species. Furthermore, the oxygen atom can
also act as a π-donor. The formation of the σ- and π-bond as
well as the donation of electrons into the empty bonding
d-orbitals of silver lower the energy level and stabilize the
metal centre, thereby increasing its catalytic activity.
3
1,32
estimated based on the mean crystallite size.
The number
of surface atoms available for reaction decreased with increas-
ing silver content (Table S1†). This decrease is most drastic
when the metal loading changed from 5 to 10 wt% where an
eight-fold increase in the crystallite size from 5 to 40 nm was
observed. However, the TOF based on the number of silver
−
1
atoms at the surface increased from 19.6 h for 5 wt%
−
1
Ag/Al O to a maximum of 216 h
for the 10 wt%
2
3
Unexpectedly, hydrogen pretreatment had a negative effect
on the activity of the catalyst. After reduction of the 10 wt%
Ag/Al O catalyst at 300 °C under flowing hydrogen for 0.5 h,
Ag/Al
2
O
3
catalyst (Fig. 2). With even higher loading of 15
and 20 wt% Ag, the TOF decreased moderately to 173 and
125 h , respectively. It finally reached a constant value of
about 90 h
−
1
2
3
−
1
the initial rate of reaction was low, and the conversion after
h in air was only 10% as compared to 75% for the unreduced
for the silver powders which had much
1
larger particle sizes (2.2–22 μm) than the supported
catalyst (Fig. 1). The better activity of the latter suggests that
Ag ions are important for decarboxylation. Indeed, the con-
catalysts. The kinetic profiles show clearly that the supported
+
catalyst (10 wt% Ag/Al
for the protodecarboxylation than any of the silver powders or
bulk Ag O (Fig. 3). The initial rate for the Ag O catalysts was at
least 9 to 15 times smaller than that of 10 wt% Ag/Al . Thus,
2 3
O ) has a far superior catalytic activity
version over the reduced catalyst increased slowly over time
to reach 86% after 5 h (Table 3, entry 14). As the reaction was
conducted in a system open to the environment, diffusion of
air could have led to reoxidation of the metallic silver catalyst
and consequently, the formation of a more active catalyst.
2
2
2 3
O
the supported catalyst does not only gain from the higher
dispersion of the metal, but also shows some additional
activation due to the support.
2
Indeed, when the reaction was conducted under N using the
reduced Ag catalyst, the lack of O prevented the reoxidation
of the catalyst, resulting in almost no conversion. In compari-
The results suggest that the protodecarboxylation of
2-nitrobenzoic acid is more facile over silver crystallites of
2
2
son, use of the unreduced Ag catalyst in the inert N atmo-
sphere had no adverse effect on the rate of the reaction, and
the kinetic profile was similar to that for the reaction in air
(Fig. 1).
3
.2.2 Effect of particle size on activity. The influence of
particle size on the protodecarboxylation of 2-nitrobenzoic
acid was studied at 120 °C. The quantity of catalyst was
adjusted so that the same amount of silver was used in each
reaction. The initial rate of reaction (integrated over the first
−
1
−1
hour) increased from 5.1 to 7.6 mmol mmolAg
h
for a
Fig. 2 TOF as a function of silver particle size for (○) Ag/Al
2 3
O and (■)
Ag powder. (●) denotes homogeneous catalyst AgOAc.
Fig. 1 Conversion of nitrobenzoic acid in air (open symbols) and N
closed symbols) over 10 wt% Ag/Al calcined in air at 500 °C (◇, ◆)
and after reduction in hydrogen at 300 °C for 0.5 h (□, ■).
2
Fig. 3 Kinetic profiles for protodecarboxylation of 2-nitrobenzoic acid
over (◆) 10 wt% Ag/Al , (▲) Ag O (ex-Ag CO ), (×) Ag O (commercial)
and (■) Ag powder (ex-Ag CO ).
(
2
O
3
2
O
3
2
2
3
2
2
3
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 516–523 | 519

View Article Online
Paper
Catalysis Science & Technology
Table 4 Effect of amount of K
2 3
CO on the protodecarboxylation of
bigger dimensions. As the crystallite size increases, the abun-
dance of low coordination sites, i.e., atoms at edges, steps and
corners, decreases relative to those located at terraces or
2-nitrobenzoic acid
a
Ag−1 −1
b
K
5
2
CO
3
(mol%) Initial rate (mmol mmol
h
)
Conversion (%)
3
3,34
planes.
tially at sites with higher coordination. This concept is further
supported by the higher activity of 10 wt% Ag/Al compared
Hence, the reaction appears to take place preferen-
0.80
7.64
2.65
1.20
41
93
56
53
15
O
2 3
30
50
with that of silver acetate. Silver acetate acts as a homogeneous
catalyst as it dissolves in the hot acidic reaction medium. The
a
b
Within 1 h. After 4 h.
initial rate of reaction for the heterogeneous 10 wt% Ag/Al
2 3
O
catalyst was only slightly higher than for silver acetate, 7.6 vs.
−
1
−1
6
.2 mmol mmolAg h . The difference in catalytic activity
+
nitrobenzoate is the actual substrate, or potassium ions (K )
play a role in modifying and activating the surface of the
catalyst. To further understand the role of the base, the reac-
tion was carried out starting with potassium nitrobenzoate
becomes much more significant when the comparison is made
on the number of available silver ions. For the homogenous
catalyst, this is all the silver added as silver acetate but for the
heterogeneous catalyst, it is limited only to silver atoms at the
external surface of the particles. The TOF is ~35 times higher
for the heterogeneously catalyzed reaction than for the homo-
geneous system.
.2.3 Role of the base. Various alkali metal carbonates
such as Li CO , Na CO , K CO and Cs CO were tested
under identical concentrations (15 mol% with respect to
-nitrobenzoic acid). It is assumed that the alkali carbonate
reacts with 2-nitrobenzoic acid, forming alkali 2-nitrobenzoate,
CO and water. Due to the limited amount of base (15 mol%),
(
Table 5). A limiting conversion of 27% was reached after 4 h
which can be attributed to a lack of protons within the sys-
tem that prevents the formation of nitrobenzene after decar-
boxylation. This was confirmed because the conversion
increased to 67% and 89% when the reaction was carried out
in the presence of 15 and 100 mol% acetic acid, respectively.
Hence, while the base is needed to generate the nitrobenzo-
ate, the presence of both nitrobenzoate and 2-nitrobenzoic
acid is necessary for the reaction.
3
2
3
2
3
2
3
2
3
2
2
An equivalent amount of KOH gave the same rate of reac-
tion as K CO (Fig. 5), showing that the rate was independent
2 3
of the type of base. This is not surprising as in the presence
the reaction medium still contains some free acid. The activity
for protodecarboxylation increased with the size of the alkali
metal ions in the order Li < Na < K (Fig. 4). With K
2
CO
3
, the
−1
−1
initial rate was 7.6 mmol mmolAg
h
and 93% conversion
was attained after 4 h. This may be due to an increase in base
strength as the size of alkali metal ion increases. Furthermore,
the degree of solvation is reduced for bigger cations, hence
Table 5 Protodecarboxylation of 2-nitrobenzoic acid and potassium
2
-nitrobenzoate under different conditionsa
increasing their availability for reaction. However, with Cs
2 3
CO
K CO
2 3
AcOH
Conversion
as the base, the conversion after 4 h was only 69.5%, with an
Entry Substrate
(mol%) (mol%) (%)
−1
−1
initial rate of 1.41 mmol mmolAg h . The low activity may be
due to blockage of the active sites in the Ag O layer when the
1
2
3
4
5
2-Nitrobenzoic acid
2-Nitrobenzoic acid
Potassium 2-nitrobenzoate
Potassium 2-nitrobenzoate
Potassium 2-nitrobenzoate
—
15
—
—
—
—
—
—
15
100
6
93
27
67
89
2
large cesium ion adsorbs onto the adjacent oxygen atom.
Reducing the amount of K CO from 15 to 5 mol% decreased
2
3
−
1
−1
the initial rate by a factor of ten to 0.80 mmol mmolAg h ,
and the conversion reached only 41% after 4 h (Table 4). A
concentration of 15 mol% is optimum as increasing the K CO
a
Reaction conditions: 2 mmol 2-nitrobenzoic acid, 10 wt% Ag/Al
2 3
O
(0.2 mmol Ag), 15 mol% K
2
CO , 4 mL DMA, 120 °C, 4 h.
3
2
3
amount beyond this gave lower activity (Fig. S6†).
The vast improvement in the conversion from 6 to 93%
after the addition of 15 mol% K CO suggests that either
2
3
Fig. 5 Protodecarboxylation of 2-nitrobenzoic acid over 10 wt%
Ag/Al in the presence of (◆) 0.3 mmol KCl (■) 0.3 mmol K SO
(△) 0.3 mmol K CO (○) 0.3 mmol KOH and (●) 0.6 mmol KOH.
Fig. 4 Effect of different alkali metal carbonates on initial rate of
protodecarboxylation of 2-nitrobenzoic acid.
2
O
3
2
4
2
3
520 | Catal. Sci. Technol., 2014, 4, 516–523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
of excess 2-nitrobenzoic acid, there would be no K CO or
2
3
+
KOH present. To study if the K ions play a role in the reac-
tion, a neutral salt, K SO , was used instead of the base. A
conversion of 62% was obtained after 4 h, showing that K
ions are important in the reaction. The rate of reaction is
2
4
+
lower than with K
2 3
CO , but this may be due to the limited
solubility of K SO in DMA. However, when KCl was used
2
4
Scheme 2 Ortho-substituent coordinating to the Ag centre during
instead of K SO , no protodecarboxylation of 2-nitrobenzoic
acid was observed, suggesting that the Cl reacted with the
2
4
decarboxylation.
−
Ag O overlayer of the catalyst to form unreactive AgCl.
2
decarboxylation. This inductive effect decreases with increasing
distance for substituents in meta- and para-positions.
3
.3 Scope of reaction
The scope of the reaction was explored using various
substituted aromatic carboxylic acids (Table 6). Benzoic acid
did not undergo any protodecarboxylation. This is in contrast
to the conversion of 93% observed with 2-nitrobenzoic acid
and shows that an ortho-substituent must be present for
protodecarboxylation to proceed in high yields over the hetero-
geneous Ag/Al O catalyst. Although the meta-substituted
Replacing the strongly electron withdrawing nitro group
with a bromo group led to a lower conversion of 51% (Table 6,
entry 7). Benzoic acids ortho-substituted with chloro or fluoro
groups were highly active towards protodecarboxylation
(Table 6, entries 8–10). While carboxylic acids with activating
substituents such as methyl, hydroxyl, cyano and amino
groups were inert to protodecarboxylation over our catalyst,
good activity was observed for 2-methoxybenzoic acid. This
may be due to coordination of the lone pair of electrons on
the oxygen of the methoxy group to the metal center, thus
stabilizing the intermediate formed during the decarboxyl-
ation step. Protodecarboxylation of heteroaromatic carboxylic
acids also proceeded with high conversions (Table 6, entries
17–21). These results show that our heterogeneous Ag catalyst
shows a similar performance to previously described homo-
2
3
3
-nitrobenzoic acid did undergo some protodecarboxylation,
the conversion was only 15%. When two nitro groups are
present at the meta positions, the conversion was higher, and
3
,5-dinitrobenzoic acid could be decarboxylated with 52%
yield. However, para-substituted 4-nitrobenzoic acid was not
active towards protodecarboxylation. This agrees well with pre-
3
5
16
vious results by the groups of Gooßen and Larrosa who pos-
tulated that an intermediate is formed between the carboxylate
group, the ortho-substituent and the metal center (Scheme 2).
A similar complex can also form between an adsorbed
1
5–17
geneous Ag catalysts,
where both can tolerate a range of
substituents such as nitro, methoxy, bromo, chloro, fluoro
groups as well as heteroaromatic carboxylic acids.
+
benzoate anion and surface Ag sites on the heterogeneous
catalyst. An electron-withdrawing group at the ortho position
can exert an inductive effect via the σ-backbone, favouring
The reuse of the Al
ied. After the batch reaction, the catalyst was removed from
O -supported silver catalyst was stud-
2 3
2 3
Table 6 Protodecarboxylation of various aromatic carboxylic acids over 10 wt% Ag/Al O
a
Entry
Substrate
Position of substituent
Effect of substituent
Conv. (%)
Sel. (%)
1
2
3
4
5
6
7
8
9
Benzoic acid
2-Nitrobenzoic acid
3-Nitrobenzoic acid
4-Nitrobenzoic acid
—
—
0
93
15
0
—
b
Ortho
Meta
Para
Di-meta
Ortho
Ortho
Ortho
Ortho
Ortho
Ortho
Ortho
Ortho
Ortho
Ortho
Ortho
—
Strongly deactivating
Strongly deactivating
Strongly deactivating
Strongly deactivating
Strongly deactivating
Weakly deactivating
Weakly deactivating
Weakly deactivating
Weakly deactivating
Weakly activating
Weakly activating
Strongly activating
Strongly activating
Strongly activating
Strongly activating
—
100
100
—
100
100
100
100
100
100
—
3,5-Dinitrobenzoic acid
4-Fluoro-2-trifluoromethylbenzoic acid
2-Bromobenzoic acid
2-Chlorobenzoic acid
2-Chloro-4-nitrobenzoic acid
2-Fluoro-4-chlorobenzoic acid
2-Toluic acid
2-Phenyl-benzoic acid
2-Methoxybenzoic acid
Salicylic acid
2-Cyanobenzoic acid
Anthranilic acid
2-Picolinic acid
2-Thiophenecarboxylic acid
2-Furoic acid
2-Benzothiophene carboxylic acid
2-Benzofuran carboxylic acid
52
87
51
89
99
98
0
0
89
0
10
11
12
13
14
15
16
17
18
19
20
21
—
46
—
—
0
0
—
c
c
c
88
100
98
98
100
100
100
100
100
100
—
—
—
—
—
—
—
—
a
b
c
Reaction conditions: 2 mmol substrate, 10 wt% Ag/Al
O
2 3
(0.2 mmol Ag), 15 mol% K
2
CO
3
, 4 mL DMA, 120 °C. After 16 and. 4 h. Carried
out in an autoclave as products have low boiling points.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 516–523 | 521

View Article Online
Paper
Catalysis Science & Technology
the reaction mixture through centrifugation, and washed
with DMA followed by water to remove any adsorbed organics
and base residues, respectively. The recovered catalyst was
dried in vacuum and recalcined at 500 °C for 1 h. The test
results showed no significant loss in activity for the protode-
carboxylation of 2-nitrobenzoic acid (Table 3, entry 15). The
X-ray diffractogram of the recycled catalyst revealed no
changes in the silver phase or Al O support, indicating the
anion leads to a surface-bound negatively charged aryl–silver
intermediate and the release of CO . A second incoming
2
2-nitrobenzoic acid molecule interacts with the neighbouring
potassium ion via the oxygen of its carboxylic acid group,
aiding in the deprotonation step. The released proton binds to
the aryl–silver intermediate and forms the product, nitroben-
zene. Upon desorption of nitrobenzene, the 2-nitrobenzoate
anion adsorbs onto the catalyst surface for the next catalytic
cycle. Hence, only a catalytic amount of the base is needed for
the initial formation of the nitrobenzoate anion. The base can
also be substituted by a neutral salt, e.g., potassium sulphate,
where the potassium ion facilitates the deprotonation of the
2
3
stability of the alumina-supported Ag catalyst under the reac-
tion conditions. Hence, this simple catalytic system offers
ease of preparation, handling and reuse without needing
2 3
other additives aside from K CO for activity.
2-nitrobenzoic acid.
3
.4 Reaction mechanism
4. Conclusion
Based on these results, a mechanism for the protodecarboxylation
Silver supported on alumina and silica formed active cata-
lysts for the protodecarboxylation of 2-nitrobenzoic acid. The
activity for protodecarboxylation of 2-nitrobenzoic acid was
reaction is proposed (Scheme 3). In the presence of the base, e.g.,
K CO , the free carboxylic acid is transformed to potassium
2 3
nitrobenzoate. The observation that relatively large catalyst
particles are required for good activity suggests that the sub-
2 3
maximal for 10 wt% Ag/Al O . The TOF for this catalyst,
+
which contains silver particles of about 40 nm size, was
strate interacts with more than one Ag site. The nitrobenzoate
−
1
2
16 h , which is much higher than that for the homoge-
anion contains carboxylate and nitro groups which are electro-
−
1
+
neous silver acetate catalyst (TOF 6.2 h ) and also higher
than metallic silver powder (90 h ). XPS revealed that the
silver catalyst was covered by a surface layer of silver oxide,
statically attracted to the positively charged Ag ions at the
−
1
2
Ag O surface layer. Decarboxylation of the 2-nitrobenzoate
2
Ag O. The silver oxide layer, polar aprotic solvents such as
DMSO, DMA or NMP, and a base are all required for high
activity. In the presence of the base, 2-nitrobenzoic acid
forms 2-nitrobenzoate. The presence of both acid and salt is
required for reaction. Mechanistic studies indicate that the
catalytic cycle involves the interaction of substrate molecules
+
with more than one Ag site. This explains the need for an
extended catalytic site, and the fact that moderately large
metal particles (ca. 40 nm) give the highest catalytic activity.
Acknowledgements
Financial support from National University of Singapore
under grant numbers R-143-000-418-112 and R-143-000-476-112
is gratefully acknowledged.
Notes and references
1
L. J. Goossen, F. Collet and K. Goossen, Isr. J. Chem.,
010, 50, 617.
N. Rodriguez and L. J. Goossen, Chem. Soc. Rev.,
011, 40, 5030.
2
2
2
3
4
R. Shang and L. Liu, Sci. China: Chem., 2011, 54, 1670.
A. F. Shepard, N. R. Winslow and J. R. Johnson, J. Am. Chem.
Soc., 1930, 52, 2083.
5
6
M. Nilsson, Acta Chem. Scand., 1966, 20, 423.
A. Cairncross, J. R. Roland, R. M. Henderson and
W. A. Sheppard, J. Am. Chem. Soc., 1970, 92, 3187.
T. Cohen, R. W. Berninger and J. T. Wood, J. Org. Chem.,
7
8
1
978, 43, 837.
L. J. Goossen, W. R. Thiel, N. Rodriguez, C. Linder and
Scheme 3 Proposed catalytic cycle for the protodecarboxylation of
ortho-substituted benzoic acid over an oxidized silver surface.
B. Melzer, Adv. Synth. Catal., 2007, 349, 2241.
522 | Catal. Sci. Technol., 2014, 4, 516–523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
9
L. J. Goossen, F. Manjolinho, B. A. Khan and N. Rodriguez,
J. Org. Chem., 2009, 74, 2620.
23 K. Shimizu, K. Shimura, M. Nishimura and A. Satsuma,
ChemCatChem, 2011, 3, 1755.
1
1
1
1
1
1
1
1
1
1
0 M. Rudzki, A. Alcalde-Aragonés, W. I. Dzik, N. Rodríguez
and L. J. Gooßen, Synthesis, 2012, 184.
1 J. S. Dickstein, C. A. Mulrooney, E. M. O'Brien, B. J. Morgan
and M. C. Kozlowski, Org. Lett., 2007, 9, 2441.
24 G. J. K. Acres, A. J. Bird, J. W. Jenkins and F. King,
in Catalysis, ed. C. Kemball and D. A. Dowden, The Royal
Society of Chemistry, London, 1981, vol. 4, pp. 1–30.
25 A. L. Patterson, Phys. Rev., 1939, 56, 978.
26 M. Richter, M. Langpape, S. Kolf, G. Grubert, R. Eckelt,
J. Radnik, M. Schneider, M. M. Pohl and R. Fricke, Appl.
Catal., B, 2002, 36, 261.
27 G. B. Hoflund and Z. F. Hazos, Phys. Rev. B: Condens. Matter
Mater. Phys., 2000, 62, 11126.
28 S. W. Gaarenstroom and N. Winograd, J. Chem. Phys.,
1977, 67, 3500.
2 Z. M. Sun and P. Zhao, Angew. Chem., Int. Ed.,
2
009, 48, 6726.
3 S. Dupuy, F. Lazreg, A. M. Z. Slawin, C. S. J. Cazin and
S. P. Nolan, Chem. Commun., 2011, 47, 5455.
4 J. Cornella, M. Rosillo-Lopez and I. Larrosa, Adv. Synth.
Catal., 2011, 353, 1359.
5 L. J. Goossen, C. Linder, N. Rodriguez, P. P. Lange and
A. Fromm, Chem. Commun., 2009, 7173.
6 J. Cornella, C. Sanchez, D. Banawa and I. Larrosa, Chem.
Commun., 2009, 7176.
29 C. D. Wagner and A. Joshi, J. Electron Spectrosc. Relat.
Phenom., 1988, 47, 283.
30 D. Briggs, in: Handbook of X-ray Photoelectron Spectroscopy
ed. C. D. Wanger, W. M. Riggs, L. E. Davis, J. F. Moulder
and G. E. Muilenberg, Perkin-Elmer Corp., Physical
Electronics Division, Eden Prairie, Minnesota, USA, 1979,
pp. 120–121.
31 A. Abad, A. Corma and H. García, Chem.–Eur. J.,
2008, 14(1), 212.
32 R. E. Benfield, J. Chem. Soc., Faraday Trans., 1992, 88, 1107.
33 A. M. Doyle, S. K. Shaikhutdinov and H.-J. Freund, Angew.
Chem., Int. Ed., 2005, 44, 629.
7 P. Lu, C. Sanchez, J. Cornella and I. Larrosa, Org. Lett.,
2
009, 11, 5710.
8 J. Hagen, Industrial Catalysis, Wiley-VCH, Weinheim, 2006,
pp. 99–222.
9 A. M. Ruppert and B. M. Weckhuysen, in Handbook of
Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth,
and J. Weitkamp, Wiley-VCH, Weinheim, 2008, vol. 3, pp.
1
178–1188.
2
2
0 A. T. Bell, Science, 2003, 299, 1688.
1 H. Einaga and A. Ogata, Environ. Sci. Technol.,
34 A. Corma, P. Serna, P. Concepción and J. J. Calvino, J. Am.
Chem. Soc., 2008, 130, 8748.
2
010, 44, 2612.
2
2 K.-I. Shimizu, M. Tsuzuki, K. Kato, S. Yokota, K. Okumura
35 L. J. Gooßen, N. Rodríguez, C. Linder, P. P. Lange and
A. Fromm, ChemCatChem, 2010, 2, 430.
and A. Satsuma, J. Phys. Chem. C, 2006, 111, 950.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 516–523 | 523